Simultaneous neuronal recordings and neuromodulation were enabled by multifunctional fiber neurotechnology. This dataset contains neurophysiological and behavioral data recorded from a macaque performing a working memory task. Recordings were performed in the premotor cortex or putamen. During the recordings, there was intracranial neuromodulation.

There are 5 recording types and corresponding files

• pmc_gaba: Recordings in the premotor cortex with intracranial GABA infusion(s) (n = 4)
• pmc_sal: Recordings in the premotor cortex with intracranial saline infusion(s) (n = 3)
• pmc_acsf: Recordings in the premotor cortex with intracranial aCSF infusion(s) (n = 2)
• pmc_noic: Recordings in the premotor cortex with no intracranial infusion (n = 4)
• put_GABA: Recording in the putamen with intracranial GABA infusion
• pmc_noic: Recordings in the putamen with no intracranial infusion.

Device data is stored in device_data.zip

Additionally, we provide preprocessed data in experimental_data.zip and processed_data.zip

Further detail regarding experimental details can be found in the associated manuscript (Garwood, et al.)

Description of experimental data and file structure

Each data file contains 2 variables and 4 structures:
Variables:

• fs_lfp: The sampling rate of local field potential (lfp) recordings
• fs_spike: The sampling rate of single unit activity recordings Structures:
• infusion (when applicable; empty otherwise): Data related to the intracranial infusions
  • start: The start time(s) of intracranial infusion(s) in seconds, relative to the beginning of the recording
  • end: The end time(s) of intracranial infusion(s) in seconds, relative to the beginning of the recording
  • rate: The infusion rate(s) of intracranial infusion(s) in nL/min
  • drug: String corresponding to the drug infused
• lfp: Data related to local field potential oscillations
  • lfp: 4xN matrix of LFP data
  • lfp_elec: The LFP electrode used for subsequent analysis
• sua: Data related to single unit activity
  • unit_locs: Indices, relative to the beginning of the recording, corresponding to timepoints when each (sorted) unit fired an action potential
  • unit_spikes: The corresponding (unsorted) event index for each sorted event, relative to the first event index
  • spike_locs_all: The indices of all valid spike events, relative to the beginning of the recording
  • spikes_all: The (unsorted) waveforms of all spike events, in chronological order
  • spike_length: The duration of each spike waveform (in ms)
  • spike_threshold: The threshold used to define spike events, relative to the standard deviation of the raw spike data at baseline
  • noise_thresh: The amplitude beyond which threshold crossings were considered to be an artifact (in uV)
• task_info: Data related to the behavioral task
  • fs_task: Sampling rate used to convert trial indices to time (in sec)
  • trials: Trial start and end indices, relative to the beginning of the recording
  • samples: Start and stop indices corresponding to when the sample was visible on the screen
  • matches: Start and stop indices corresponding to when the match was visible on the screen.
  • correct_trials: Binary variable, where 0 indicates incorrect trials, 1 indicates correct trials
  • complete_trials: Binary variable, where 0 indicates incomplete trials, 1 indicates complete trials
  • no_fix: Binary variable, where 0 indicates successful fixation, 1 indicates unsuccessful fixation
  • sample_id: The ID of the sample (1-3)
  • correct_loc: The location of the correct selection
  • reaction time: The time between when the match appeared and when a selection was made (in s)

Note that one aCSF file was included in behavioral analysis but not electrophysiology analysis, because the infusion rate + volume exceeded that of the GABA recordings.

We additionally provide the output of fooof analysis in experimental_data.zip/data_for_fooof

• The data files have the following naming convention: fooof_output_exptype_6.mat where 6 corresponds to the FOOOF parameter for the max number of spectral peaks in each spectra
• Each data file contains fooof parameters estimated from 2N spectral samples, where samples 1:N occurred before intracranial infusions and samples N+1:2N occurred after intracranial infusions (Garwood 2023, Methods)
• Each data file contains 6 variables:
  • aperiodic_signal: two columns corresponding to the aperiodic exponent and aperiodic offset across 2N samples
  • fooof_spectrum: 2Nxn_freq matrix corresponding to the estimated spectra of 2N samples
  • fooof_freq: n_freq vector of frequencies corresponding to the columns in fooof_spectrum (where frequencies between 59.5 and 60.5 Hz have been removed)
  • peak_cf: 2Nx6 matrix of center frequencies across up to 6 spectral peaks; if peak_cf(n,p+1:6) = 0, only p spectral peaks were identified in that sample
  • peak_pow: 2Nx6 matrix of power across up to 6 spectral peaks; if peak_pow(n,p+1:6) = 0, only p spectral peaks were identified in that sample
  • session_tags: 2Nx1 vector containing the session number corresponding to each spectral sample (also saved in the files session_tags_exptype.mat)
• freqs.csv
  • contains the frequencies of the original spectral samples

Description of device data and file structure

Device data is organized into three subfolders: DMA, fluidic, and impedance.

• DMA: Dynamic material analysis data
  • Contains DMA data for three fiber samples and one steel sample in .csv format.
• fluidic: fluidic rate characterization data
  • Contains 30 .mat files in the format of fluidic infusion data from 10-100 nl/min
    • Naming convention: rate_nlm_trial_x.mat where "nlm" refers to nanoliters/minute as the units of time
    • Each file contains two variables: time, and volume (cumulative volume infused over time).
  • Subfolder 'raw_videos' contains the video recordings corresponding to each trial in the parent directory.
• impedance: impedance spectroscopy data
  • Contains 24 .txt files containing impedance spectroscopy data (frequency vs. impedance)
    • Naming convention: devx_elecy.txt or devx_postAC_elecy.txt (postAC indicates spectroscopy performed after autoclave sterilization)

Description of preprocessed data and file structure

Processed data folder includes spike sorting results and example SS-GLM and AR models

• spike_data: folder containing spike sorting results
  • Naming convention: spikes_cluster_location_exptypex.mat where 'x' indicates that experiment number
  • Each file contains 3 variables:
    • s_time: experiment time sampled at 30 kHz
    • spikes_cluster: the spike cluster id of every threshold crossing
    • spikes_locs: n_units by 1 cell array containing the index of every spike of the corresponding unit. Single unit times are given by s_time(spike_locs{unit_num})
• SSGLM models follow the naming convention location_exptypex_unity_trialvariant_Rz.mat where 'x' indicates the experiment number, 'y' indicates the unit id, and 'z' indicates the model order.
  • Additional subscripts include '_nohist.mat' indicating the model has no history terms and '_stationary.mat', indicating that the parameters are stationary across trials.
  • Each file contains estimated SSGLM parameters (Garwood 2023, Methods, README_code.md). Each parameter is an 1 by M cell array, where M corresponds to the number of models estimated for the corresponding trial variant.
    • gammahat: estimated values of gamma (history coefficients) on the final EM iteration
    • xK: the expected value of the state (effect of task subphase on firing rate in a given trial)
    • gammahatall: estimated values of gamma across all EM iterations
    • logll: log likelihood of the model
    • nIter: number of EM iterations performed
    • Qhat: estimated state noise variance (state noise covariance is assumed to be diagonal) on the final EM iteration
    • Qhat: estimated state noise variance across all EM iterations
    • WK: state covariance within trials\, W_(k|K)
    • WKu: state covariance across trials\, W_(k\,u|K)
    • stimCIs: optional SSGLM output; confidence intervals for state estimates derived with Monte Carlo; computationally intensive to compute, currently empty
    • stimulus: optional SSGLM output; median of the state estimates derived with Monte Carlo; computationally intensive to compute, currently empty
    • fitResults: optional output; extended fitResults if this is enabled in the nSTAT function, DecodingAlgorithms_IG.m; currently empty to avoid unnecessary data storage.
• AR models follow the naming convention location_exptypex_lfp_trialvariant_Rz.mat
  • Each file contains estimated AR parameters with and without task subphase covariates (variable names _with_encoding and _without_encoding). Each parameter is an 1 by M cell array, where M corresponds to the number of models estimated for the corresponding trial variant.
    • E: AR model residuals
    • V: AR model conditional variance
    • LL: AR model log likelihood
    • EstMdl: Matlab ARIMA structure for the estimated AR model

Code/Software

Code will be available in a permanently archived repository located at https://github.com/igarwood/NHP_fibers

See the main README_code.md file in https://github.com/igarwood/NHP_fibers

MIT license.